Single-blind controlled release testing to evaluate the performance of an in-situ methane detection system

Abstract

The accurate detection and quantification of Methane (CH4) emissions from equipment at oil and gas facilities is critical for resolving environmental concerns associated with Greenhouse Gas (GHG). This study presents the evaluation of a proprietary CH4 detection solution deployed at a simulated oil and gas wellhead configuration on a green field site in south Texas. In this study, an automated controlled release rig was developed and used to conduct a single-blind, controlled release test program using a protocol developed specifically for the solution under test. The test program lasted for 57 days and included 5,153 discrete 15-minute reporting intervals with release rates ranging from 0.5 kgh-1 to 37.5 kgh-1, and release durations ranging from 15 minutes to 3 hours. Upon completion of testing, a binary classification scheme was used to evaluate the performance of the solution by comparing the reports from the solution to the ground-truth emissions generated by the controlled release rig. For controlled releases greater than 2.5 kgh-1, the solution showed a true positive rate (TPR) of 86 %, and a false negative rate (FNR) of 14 %. For release rates less than 2.5 kgh-1 the solution showed true negative rate (TNR) of 99.8 %, and a false positive rate (FPR) of 0.2 %. The solution showed a 90 % probability of reporting emissions greater than 5.5 kgh-1 and a 50 % probability of reporting emissions greater than 4.4 kgh-1. The solution performance improved with increasing wind-speed with 90 % and 50 % probability of reporting decreasing to 3.8 kgh-1 and 3.0 kgh-1, respectively, for wind speeds greater than 2 ms-1.